REVIEW

Esophageal Perforation as a Complication of

Esophagogastroduodenoscopy

Nisha L. Bhatia, MD1 Fifty years ago, esophageal perforation was common after rigid upper endoscopy.
Joseph M. Collins, MD2 The arrival of flexible endoscopic instruments and refinement in technique have
Cuong C. Nguyen, MD3 decreased its incidence; however, esophageal perforation remains an important
Dawn E. Jaroszewski, MD4 cause of morbidity and mortality. This complication merits a high index of clinical
Holenarasipur R. Vikram, MD5 suspicion to prevent sequelae of mediastinitis and fulminant sepsis. Although the
Joseph C. Charles, MD1 risk of perforation with esophagogastroduodenoscopy alone is only 0.03%, this risk
can increase to 17% with therapeutic interventions in the setting of underlying
1
Division of Hospital Internal Medicine, Mayo esophageal and systemic diseases. A wide spectrum of management options exist,
Clinic Arizona, Phoenix, Arizona ranging from conservative treatment to surgical intervention. Prompt recognition
2
Division of Radiology, Mayo Clinic Arizona, Phoe- and management, within 24 hours of perforation, is critical for favorable outcomes.
nix, Arizona Journal of Hospital Medicine 2008;3:256 –262.
3 © 2008 Society of Hospital Medicine.
Division of Gastroenterology, Mayo Clinic Ari-
zona, Phoenix, Arizona
4
Division of Cardiothoracic Surgery, Mayo Clinic KEYWORDS: esophagogastroduodenoscopy, esophageal perforation, mediastinitis,
Arizona, Phoenix, Arizona sepsis, endoscopy.
5
Division of Infectious Diseases, Mayo Clinic Ari-
zona, Phoenix, Arizona

E sophagogastroduodenoscopy (EGD) carries a small but serious

risk of esophageal perforation.1–3 With its potential for sepsis
and fatal mediastinitis, prompt recognition and treatment are
essential for favorable outcomes. The risk of perforation with
diagnostic flexible EGD is 0.03%, which is an improvement from
the 0.1%-0.4% risk associated with rigid endoscopy.4 However, the
risk of perforation can dramatically increase to 17% depending on
the methods of therapeutic intervention and underlying risk fac-
tors (Table 1).1,5–7
It is estimated that 33%-75% of all esophageal perforations are
iatrogenic.8 Of those caused by EGD, therapeutic interventions
portend an increased risk compared with the risk of diagnostic
endoscopy alone (Table 2).4 With the expanding role of flexible
EGD and the increasing number of procedures performed, this
modest risk per procedure still translates into a sizable number of
perforations with their ensuing complications.4,7 Mortality rates
following esophageal perforation may approach 25%.9

ANATOMY AND PATHOPHYSIOLOGY

The most common site of perforation is at the level of the crico-
pharyngeus, as it is a narrow introitus leading to the esophagus.
The risk of perforation at this location is further increased with the
presence of a Zenker’s diverticulum or cervical osteophytes. The
second most common site is proximal to the lower esophageal
sphincter because of the angulation of the hiatus and the high
frequency of esophageal webs, rings, reflux strictures, and hiatal

256 © 2008 Society of Hospital Medicine

DOI 10.1002/jhm.289
Published online in Wiley InterScience (www.interscience.wiley.com).
TABLE 1 agus is compressed over the underlying spinal col-
Risk Factors for Esophageal Perforation umn. Thoracic perforations, however, are more com-
monly seen with organic esophageal obstruction.
Level of operator experience
Underlying esophageal disease These obstructions may be caused by an underlying
Zenker’s diverticulum inflammatory process, benign stricture, or neoplasm.
Eosinophilic esophagitis In these cases, the risk of thoracic perforation is in-
Esophageal or mediastinal irradiation creased with blind procedures. Thoracic perforations
Esophageal malignancy
carry a worse prognosis if diagnosis is delayed, or if
Esophageal strictures
Systemic disease the underlying obstruction cannot be removed.10
Anterior cervical osteophytes Esophageal perforation leads to periesophageal
Advanced liver cirrhosis tissues being contaminated by food, secretions, air,
Diabetes mellitus or gastric contents and may be followed by chem-
Scleroderma
ical tissue injury and infection. The nature and
Complexity of intervention
Esophageal stent placement extent of infection depend on the site of esophageal
Pneumatic dilation perforation. Cervical esophageal perforation can
Other cause retropharyngeal space infection, which has
Heavy sedation the potential to extend directly into the posterior
Advanced age
mediastinum via the “danger” space, which is be-
Data from Clouse,1 Sorbi et al.,5 Mandelstam et al.,6 and Hernandez et al.7 tween the retropharyngeal and prevertebral spaces
and extends from the base of the skull descending
freely throughout the entire length of the posterior
TABLE 2
mediastinum. With thoracic perforations, esopha-
Risk of Esophageal Perforation in Diagnostic and Therapeutic
Esophagogastroduodenoscopy geal contents can enter the pleural space by nega-
tive intrathoracic pressure with subsequent pleural
Endoscopic procedure Esophageal perforation risk contamination and empyema.8,11–13
Pathogens responsible for infections after esoph-
Diagnostic 0.03%
ageal perforation vary based on several factors includ-
Dilation 0.25% (normal esophagus)
4%-7% (achalasia)* ing site of perforation, clinical status of patient when
7% (gastric outlet obstruction)* perforation occurs (hospitalized versus not hospital-
17% (strictures due to caustic agent) ized, critically ill versus “healthy”), receipt of enteral
Thermal method (treatment of nutrition, gastric acid suppression with H2-receptor
malignancy) 10%†
antagonists or proton-pump inhibitors, immunosup-
Endoprosthesis 3%‡
Variceal sclerotherapy 1%-5% (acute perforation) pression, and recent (or current) receipt of antimicro-
2%-5% (delayed perforation) bials. In nonintubated, healthy adults not on antimi-
Band ligation 0.7% (perforation) crobial therapy, organisms in the upper esophagus
Nonvariceal hemostasis (use of 0%-2% (first hemostasis) are essentially identical to those in the oropharynx
sclerosant or cautery) 4% (hemostasis repeated within 24-48 hours)
and include viridans streptococci, Haemophilus spe-
*With dilation ⬎ 15 mm. cies, and anaerobes. During critical illness and follow-
†
Combined rate of perforation and/or fistulae, or both. ing antibiotic therapy, the normal oral flora is rapidly
‡
Combined rate of perforation, hemorrhage, and/or aspiration. replaced by aerobic Gram-negative bacilli, Staphylo-
Data from Newcomer et al.3 and Eisen et al.9 coccus aureus, and yeast.14 The stomach, which is
normally devoid of bacteria, can likewise be colonized
with pathogenic organisms in the setting of gastric
hernias. The relatively straight middle esophagus is acid suppression and enteral nutrition.15,16
an uncommon site for perforations.
Cervical perforations are less commonly caused
by organic lesions of the esophagus. Often, they are SIGNS AND SYMPTOMS
the result of technique and manipulation of the en- Esophageal perforation should be considered after
doscope, or of certain conditions associated with the EGD, dilation, sclerotherapy, variceal banding, and
jaw, neck, or spinal column that are unfavorable for esophageal stenting. However, perforation can also
endoscopy. The risk of cervical perforation increases result from other invasive procedures such as in-
with the presence of bony spurs, as the upper esoph- sertion of feeding and nasogastric tubes, rapid se-

Esophageal Perforation, Complication of EGD / Bhatia et al. 257

TABLE 3
Symptoms and Signs of Esophageal Perforation

Location of
perforation Symptom Sign*

Cervical esophagus Muscle spasm Anterior neck tenderness

Dysphonia Tenderness on cervical motion
Hoarseness Subcutaneous emphysema
Dysphagia
Thoracic esophagus Substernal chest pain Cyanosis, Dyspnea
Dysphagia Hamman’s sign†
Odynophagia Pleural effusion
Subcutaneous emphysema
Intraabdominal Epigastric pain Acute abdomen
esophagus Subcutaneous emphysema

*Patient can present with fever, sepsis, and/or shock regardless of perforation site.
†
An audible crunch with chest auscultation that may vary with the cardiac cycle; this finding is
associated with mediastinal emphysema. Data from Duncan and Wong.8

quence intubation, and transesophageal echocardi-

ography.
The clinical triad of esophageal perforation in-
cludes pain, fever, and subcutaneous air.17 In a
study by Wychulis et al., among 33 patients with
esophageal perforation, 75% demonstrated all 3
findings.10 Pain is the most sensitive finding and
occurs in nearly all patients identified with esoph-
ageal perforation. Crepitation, which results from
air dissecting along soft tissue planes of the medi-
astinum and into the neck, occurs in up to 70% with FIGURE 1. CT scan of the neck with dilute Gastrografin姞 demonstrating air
cervical perforation and 30% with thoracic perfora- (arrows) in the “danger space.”
tion.8,10,18
Clinical presentation and outcomes vary de-
pending on the location of the perforation (Table DIAGNOSIS
3).8 Cervical perforation is usually associated with Clinical suspicion of esophageal perforation should
anterior neck pain, located at the anterior border of prompt necessary radiographic studies to establish
the sternocleidomastoid muscle. Movement of the the diagnosis.18,20 Contrast-enhanced computed
neck and palpation typically aggravate the pain. tomography (CT) scans of the neck and chest are
Thoracic perforation typically presents as subster- preferable because of their increased sensitivity in
nal chest pain, often with a component of pleurisy. localizing the site and showing the extent of perfo-
Pleural effusions are present in 50% of thoracic ration and abscess. CT scans may reveal subcuta-
perforations, and mediastinitis is more likely to oc- neous or mediastinal air, abscess cavities adjacent
cur.19 Hamman’s sign, a finding characterized by to the esophagus, and fistulas between the esoph-
an audible crunch with chest auscultation, is sug- agus and mediastinum (Figs. 1 and 2).20 –22 Results
gestive of mediastinal emphysema. Perforation of of contrast studies may be negative and warrant
the intra-abdominal esophagus can result in epi- repeating within several hours.19
gastric pain and signs of acute abdomen.10,17 Sub- If CT scans cannot be performed, neck (soft-
cutaneous emphysema occurs more frequently tissue) and chest x-rays may be useful. Although
with cervical perforation but can be present regard- plain films have limited value in evaluating the
less of location.10 Secondary infections following retropharyngeal space, they can reveal soft-tissue
esophageal perforation can manifest with an accel- emphysema, a widened mediastinum, pulmonary
erated clinical course leading to sepsis and shock. infiltrates or effusions, neck abscess, and mediasti-

258 Journal of Hospital Medicine Vol 3 / No 3 / May/June 2008


FIGURE 2. CT scan of the neck with dilute Gastrografin姞 demonstrating
periesophageal air leaks (arrowheads) and extravasated contrast (arrow), con-
firming and localizing esophageal perforation.
FIGURE 3. Gastrografin姞 swallow evaluation showing extravasation (arrow)
from cervical esophagus.
nal air-fluid levels. In cervical perforation, a lateral
film of the neck can show air in deep cervical tissue
before clinical signs are apparent.
Swallow studies with Gastrografin威 (meglumine
diatrizoate) are useful in defining the exact location
of the perforation (Fig. 3). However, the false-neg-
ative rate of swallow studies can exceed 10%, espe-
cially if the patient is upright during the study.
When the contrast propagates past the site of per-
foration too quickly, it may not extravasate.23 Al-
Esophageal Perforation, Complication of EGD / Bhatia et al.
though barium may provide slightly greater con-
trast, it may add to the problem of foreign body
reaction in the area of perforation.18 An additional
complication of barium is that once it has extrava-
sated, it is not readily absorbed. The persistence of
extravasated barium makes it difficult to assess the
resolution of an esophageal tear on subsequent flu-
oroscopic or CT exams. Hence, our institution
avoids using barium to evaluate esophageal perfo-
ration, unless Gastrografin威 swallow has excluded
any major esophageal perforation. Barium swallow
may then be used to exclude small mural tears.
Some medical centers elect to routinely screen their
high-risk patients with swallow evaluations after an
EGD, although this is not common practice.8,24
If the above workup is negative, the use of EGD
may be considered for establishing the diagnosis if
a high index of suspicion remains. However, the
risks of EGD in this situation include extension of
the perforation, further extravasation of esophageal
contents, and difficulty with subsequent radio-
graphic studies to visualize the perforation.19
MANAGEMENT
Once the diagnosis of esophageal perforation has
been established, treatment options are individual-
ized based on the clinical scenario. Currently, there
are no established guidelines, and large random-
ized clinical trials comparing outcomes of operative
versus nonoperative management have not been
conducted (Fig. 4).25,26 Outcomes associated with
esophageal perforations depend on preoperative
clinical condition, comorbidities, location and size
of the perforation, nature of underlying esophageal
disease (if any), and time to establish the diagnosis
and initiate therapy.10 Delay in patient presentation
or diagnosis beyond 24 hours following esophageal
perforation has been associated with adverse out-
comes.18,27,28
A conservative approach is appropriate when
clinically stable patients with minimal symptoms
have well-contained, nontransmural tears. Man-
agement entails broad-spectrum antibiotics, noth-
ing by mouth, nasogastric suction, and parenteral
nutrition.24 Early surgical consultation is recom-
mended in all cases. Serial CT scanning is useful for
following the resolution of fistulas and tears. An
oral diet can be resumed when contrast or swallow
studies show no extravasation of dye. Cervical per-
forations typically fare well with this approach.26,29
Surgical therapy is recommended for patients
with large or uncontained esophageal perforations,
259
FIGURE 4. Algorithm for diagnosis and management of esophageal perforation.
†
Barium swallow may be considered if 1) no extravasation is seen on Gastrografin姞 swallow or 2) other imaging methods cotraindicated or unavailable.
§
Luminal pressure proximal to area of high-grade stenosis increases risk of complications with proximal esophageal perforation.

mediastinal abscesses, and/or sepsis.25,27 Surgical
options include esophageal diversion, esophagec-
tomy, or drainage with or without primary repair.
Drainage with primary repair is considered the
treatment of choice, regardless of time to presenta-
tion. Esophagectomy is considered in cases of de-
layed or neglected perforations, extensive transmu-
ral necrosis or underlying cancer.30 Operative
mortality is 0%-30% when treated within 24 hours.
This rate increases to 26%-64% when treatment is
delayed beyond 24 hours, reaffirming the impor-
tance of making a prompt diagnosis.8
Endoscopic intervention is gaining recognition
for its role in the management of esophageal per-
forations, especially when the risks make surgery
prohibitive. Therapeutic options include stenting
and clipping a perforation, as well as debriding and
draining an abscess. Endoscopists can successfully
treat traumatic nonmalignant esophageal perfora-
tions smaller than 50% to 70% of the circumference

260 Journal of Hospital Medicine Vol 3 / No 3 / May/June 2008

with self-expanding plastic stents.26 Another option
is to use metallic clipping devices to treat small
esophageal perforations (⬍1 cm).31–33 Combined
with medical management and appropriate patient
selection, the benefits of an endoscopic approach
may potentially outweigh the risks of sur-
gery.26,29,33,34
Regardless of treatment approach, the appro-
priate and timely selection of empiric antibiotic
therapy improves outcomes. Empiric antimicrobial
therapy for esophageal perforation will depend on
several host factors as well as the site of perforation.
In healthy nonhospitalized adults, ampicillin-sul-
bactam, clindamycin, and penicillin G plus metro-
nidazole are good choices because of their excellent
activity against oral microflora. In patients who are
critically ill, are hospitalized, are immunosup-
pressed, or have gastric acid suppression, initial
broad-spectrum antimicrobials such as piperacil-
lin-tazobactam, imipenem, meropenem, or a third-
generation cephalosporin plus metronidazole (or
clindamycin) should be initiated. Additional ther-
apy against methicillin-resistant Staphylococcus au-
reus or Candida sp. should be considered if the
patient is critically ill or is known to be colonized
with these organisms. Initial empiric therapy
should be modified as necessary based on culture
results. Total duration of therapy will vary based on
location and magnitude of the infection, adjunctive
surgical debridement, and pathogens involved.
SUMMARY
Despite being an extremely safe procedure, EGD
carries a known serious risk of esophageal perfora-
tion. Mortality after esophageal perforation can ap-
proach 25%. Although diagnostic endoscopy has a
perforation rate of less than 0.03%, the risk can
approach 17% with therapeutic interventions such
as stent placement and esophageal dilation. Factors
influencing the risks of perforation include proce-
dural complexity, operator experience, and under-
lying esophageal and systemic diseases. Further-
more, perforations complicated by infection can
lead to fatal mediastinitis and sepsis. The clinical
triad of esophageal perforation is fever, neck pain,
and crepitus. The optimal diagnostic study is CT
scan of the neck and thorax with water-soluble oral
contrast. Treatment options range from conserva-
tive management with broad-spectrum antibiotics
to surgery. Diagnosis of esophageal perforation
within 24 hours is essential for favorable outcomes.
Esophageal Perforation, Complication of EGD / Bhatia et al.
Address for correspondence and reprint requests: Nisha L. Bhatia, MD, Mayo
Clinic College of Medicine, Department of Internal Medicine, 5777 E. Mayo Blvd.,
Phoenix, AZ 85054; Fax: (480) 342-1388; E-mail: bhatia.nisha@mayo.edu
Received 1 May 2007; revision received 7 August 2007; accepted 4 September
2007.
REFERENCES
1. Clouse RE. Complications of endoscopic gastrointestinal di-
lation techniques. Gastrointest Endosc Clin N Am. 1996;6:
323-341.
2. Chan MF. Complications of upper gastrointestinal endos-
copy. Gastrointest Endosc Clin N Am. 1996;6:287-303.
3. Newcomer MK, Brazer SR. Complications of upper gastro-
intestinal endoscopy and their management. Gastrointest
Endosc Clin N Am. 1994;4:551-570.
4. Fernandez FF, Richter A, Freudenberg S, Wendl K, Mane-
gold BC. Treatment of endoscopic esophageal perforation.
Surg Endosc. 1999;13:962-966.
5. Sorbi D, Gostout CJ, Henry J, Lindor KD. Unsedated small-
caliber esophagogastroduodenoscopy (EGD) versus conven-
tional EGD: a comparative study [see comment]. Gastroen-
terology. 1999;117:1301-1307.
6. Mandelstam P, Sugawa C, Silvis SE, Nebel OT, Rogers BH.
Complications associated with esophagogastroduodenos-
copy and with esophageal dilation. Gastrointest Endosc.
1976;23(1):16-19.
7. Hernandez LV, Jacobson JW, Harris MS. Comparison among
the perforation rates of Maloney, balloon, and savary dila-
tion of esophageal strictures. Gastrointest Endosc. 2000;51(4
Pt 1):460-462.
8. Duncan M, Wong RK. Esophageal emergencies: things that
will wake you from a sound sleep. Gastroenterol Clin N Am.
2003;32:1035-1052.
9. Eisen GM, Baron TH, Dominitz JA, et al. Complications of
upper GI endoscopy. Gastrointest Endosc. 2002;55:784-793.
10. Wychulis AR, Fontana RS, Payne WS. Instrumental perfora-
tions of the esophagus. Dis Chest. 1969;55(3):184-189.
11. Tulman AB, Boyce HW Jr. Complications of esophageal di-
lation and guidelines for their prevention. Gastrointest En-
dosc. 1981;27:229-234.
12. Schembre DB. Infectious complications associated with gas-
trointestinal endoscopy. Gastrointest Endosc Clin N Am.
2000;10:215-232.
13. Raines DR, Branche WC, Anderson DL, Boyce HW, Jr. The
occurrence of bacteremia after esophageal dilation. Gastro-
intestinal Endoscopy 1975;22(2):86-87.
14. Safdar N, Crnich CJ, Maki DG, Safdar N, Crnich CJ, Maki DG.
The pathogenesis of ventilator-associated pneumonia: its
relevance to developing effective strategies for prevention.
Respir Care. 2005;50:725-739; discussion 39-41.
15. Torres A, El-Ebiary M, Soler N, et al. Stomach as a source of
colonization of the respiratory tract during mechanical ven-
tilation: association with ventilator-associated pneumonia.
Eur Respir J. 1996;9:1729-1735.
16. Segal R, Pogoreliuk I, Dan M, et al. Gastric microbiota in
elderly patients fed via nasogastric tubes for prolonged pe-
riods. J Hosp Infect. 2006;63(1):79-83.
17. Enat R, Levitan R. Retroesophageal abscess twenty-five days
after esophagoscopy. An unusual complication of endos-
copy. Gastrointest Endosc. 1972;18:167-168.
261
18. Sullivan M, Berry BE, Ferrante WA. The radiologist in pre-
vention and diagnosis of instrumental perforation of the
esophagus. South Med J. 1974;67:830-836.
19. Sawyer R, Phillips C, Vakil N. Short- and long-term outcome
of esophageal perforation. Gastrointest Endosc. 1995;41(2):
130-134.
20. Mizutani K, Makuuchi H, Tajima T, Mitomi T. The diagnosis
and treatment of esophageal perforations resulting from
nonmalignant causes. Surg Today. 1997;27:793-800.
21. Chong VF, Fan YF. Radiology of the retropharyngeal space.
Clin Radiol. 2000;55:740-748.
22. Endicott JN, Nelson RJ, Saraceno CA. Diagnosis and man-
agement decisions in infections of the deep fascial spaces of
the head and neck utilizing computerized tomography. La-
ryngoscope. 1982;92(6, Pt. 1):630-633.
23. DeMeester TR. Perforation of the esophagus. Ann Thorac
Surg. 1986;42:231-232.
24. Sato S, Kajiyama Y, Kuniyasu T, et al. Successfully treated
case of cervical abscess and mediastinitis due to esophageal
perforation after gastrointestinal endoscopy. Dis Esophagus.
2002;15:250-252.
25. Younes Z, Johnson DA. The spectrum of spontaneous and
iatrogenic esophageal injury: perforations, Mallory-Weiss
tears, and hematomas. J Clin Gastroenterol. 1999;29:306-
317.
26. Siersema PD. Treatment of esophageal perforations and
262 Journal of Hospital Medicine Vol 3 / No 3 / May/June 2008
anastomotic leaks: the endoscopist is stepping into the
arena. Gastrointest Endosc. 2005;61:897-900.
27. Michel L, Grillo HC, Malt RA. Operative and nonoperative
management of esophageal perforations. Ann Surg. 1981;
194(1):57-63.
28. Reeder LB, DeFilippi VJ, Ferguson MK. Current results of
therapy for esophageal perforation. Am J Surg. 1995;169:615-
617.
29. Hookey LC, Le Moine O, Deviere J. Successful endoscopic
management of a cervical pharyngeal perforation and me-
diastinal abscess. Gastrointest Endosc. 2005;61(1):158-160.
30. Port JL, Kent MS, Korst RJ, Bacchetta M, Altorki NK. Tho-
racic esophageal perforations: a decade of experience.[see
comment]. Ann Thorac Surg. 2003;75:1071-1074.
31. Seewald S, Soehendra N. Perforation: part and parcel of
endoscopic resection?[comment]. Gastrointest Endosc. 2006;
63:602-605.
32. Raymer GS, Sadana A, Campbell DB, Rowe WA. Endoscopic
clip application as an adjunct to closure of mature esopha-
geal perforation with fistulae. Clin Gastroenterol Hepatol.
2003;1(1):44-50.
33. Wewalka FW, Clodi PH, Haidinger D. Endoscopic clipping of
esophageal perforation after pneumatic dilation for achala-
sia. Endoscopy. 1995;27:608-611.
34. Egan JV, Baron TH, Adler DG, et al. Esophageal dilation.
Gastrointest Endosc. 2006;63:755-760.